News recommendation systems? purpose is to tackle the immense amount of news and offer personalized recommendations users. A major issue in capture precise representations for efficacy recommended items. Commonly, contents are filled with well-known entities different types. However, existing systems overlook exploiting external knowledge about topical relatedness among news. To cope above problem, this paper, we propose Topic-Enriched Knowledge Graph Recommendation System(TEKGR). Three...

To share and publish the domain knowledge of IoT objects, development a semantic model based directory system that manages meta-data relationships objects is required. Many researches focus on static between objects. However, because complex various resources change with time in an environment, efficient method for updating Thus, we propose IoT-DS as supports description, discovery, integration Firstly, introduce component to establish shared conceptualization. Secondly, present general...

Fog computing is a promising paradigm in terms of extending cloud to an edge network. In broad sense, fog Internet-of-Vehicles(IoV) provides low-latency services since nodes are closely located with moving cars and locally distributed. this paper, we propose architecture based on publish/subscribe model. After that, describe traffic congestion control scenario using smart light system which operates top the proposed architecture. Furthermore, upper-level domain ontology order enhance...

Knowledge graphs, which have been widely utilized in various intelligent applications, are highly incomplete. Many valid facts can be inferred from existing knowledge graphs. A promising approach for this task is a graph representation learning, aims to represent entities and relations into low-dimensional vector spaces. Most of the methods mainly focus on direct relationships between do not reflect semantics multi-hop relation paths. Although few studied problem path-based they fail...

The main focus of relational learning for knowledge graph completion (KGC) lies in exploiting rich contextual information facts. Many state-of-the-art models incorporate fact sequences, entity types, and even textual information. Unfortunately, most them do not fully take advantage structural a KG, i.e., connectivity patterns around each entity. In this paper, we propose context-aware convolutional (CACL) model which jointly learns from entities their multi-hop neighborhoods. Since directly...

Byungkook Oh, Seungmin Seo, Cheolheon Shin, Eunju Jo, Kyong-Ho Lee. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and 9th International Joint (EMNLP-IJCNLP). 2019.

Active learning attempts to maximize a task model’s performance gain by obtaining set of informative samples from an unlabeled data pool. Previous active methods usually rely on specific network architectures or task-dependent sample acquisition algorithms. Moreover, when selecting batch sample, previous works suffer insufficient diversity because they only consider the informativeness each sample. This paper proposes task-independent method using triplet loss distinguish hard in pool with...

The forthcoming smart-grid is expected to provide advanced energy services more efficiently and eco-friendly with an integrated network of various information such as a smart meter social networks. As model expressing network, knowledge graph (KG) represents formal common specification concepts relations in specific domain. It essential leverage KGs for the interoperability, reusability, scalability applications. However, existing have limitations accommodating detailed descriptions...

This paper presents an efficient method of extracting n-ary relations from multiple sentences which is called Entity-path and Discourse relation-centric Relation Extractor (EDCRE). Unlike previous approaches, the proposed focuses on entity link, consists dependency edges between entities, discourse sentences. Specifically, model two main sub-models. The first one encodes with a higher weight link while considering other attention mechanism. To consider various latent sentences, second...

The amount of RDF data being published on the Web is increasing at a massive rate. MapReduce-based distributed frameworks have become general trend in processing SPARQL queries against data. Currently, query systems that use MapReduce not been able to keep up with increase semantic annotated data, resulting non-interactive processing. principal reason intermediate results from join operations framework are so they consume all available network bandwidth. In this article, authors present an...

Both entity typing and relation extraction from text corpora are widely used to identify the semantic types of an a in knowledge graph (KG). Most existing approaches rely on pre-defined set KG. They thus cannot map mentions (relation mentions) unseen types). To fundamentally overcome limitations, we should add new entities relations KG schema. However, schema expansion traditionally requires manual conceptualization through user's observation corpus while assuming existence suitable target...

To satisfy a user’s complex requirements, Resource Description Framework (RDF) Stream Processing (RSP) systems envision the fusion of remote RDF data with semantic streams, using common models to query streams continuously. While streaming are changing at high rate and pushed into RSP systems, retrieved from different sources. With growth SPARQL endpoints that provide access data, can easily integrate streams. Such integration provides new opportunities for mixing static (or quasi-static) on...

With the rapid deployment of a number sensors, it is crucial to efficiently manage their data streams with heterogeneous properties. To achieve various sensor applications such as discovery and mashup, method retrieving meaningful information from raw required. However, hard analyze represent since sensors generate streaming different patterns continuously transmit observations servers in real-time. In this paper, we propose processing architecture retrieve data. particular, adopt machine...

Although the emergence of SPARQL endpoints that allow end-users and applications to query RDF data they want, continuous processing building a very large over diverse requires sophisticated method. However, current Stream Processing (RSP) are limited in terms scalability administrative autonomy, due their tight-coupled sources (e.g., streams) being unable coordinate with existing engines. In this paper, we propose novel continous is equipped proactive adaptation for enhancing planbased...